ComfyUI

Creator seungho__yeo ( IG ) built a morphing transition workflow on ComfyUI Cloud + Seedance 2 → seamless frame blending → character consistency preserved → cinematic This is pretty sick!

Gemini Omni Flash can swap objects, change environments, and make targeted edits to existing videos using simple text prompts. Rather than re-creating an entire clip, it can modify specific elements of a scene, making it useful for quick iterations, creative edits, and experimenting with different ideas. Want to try this workflow yourself? Link in the comments below. https://lnkd.in/d5PUrEH9

Nano Banana 2 Lite is now in Comfy. The fastest, cheapest model in the Nano Banana family. → 4-second image gen → $0.034 per 1K image If your pipeline runs on iteration volume, this changes the game. Generate 50 concept variations before lunch. Spin up ad asset batches without watching a progress bar. Prototype faster than your brief can change. Click the link below to get the workflows 👇 Nano Banana 2 Lite: Image Edit https://lnkd.in/gaiimfH3 Nano Banana 2 Lite: Text To Image https://lnkd.in/grTMVhsx

Gemini Omni Flash is now available in ComfyU via partner nodes. Google's latest video model replaces Veo entirely. Google's state of the art model with physics and video editing capabilities is now inside the world's most strongest generative AI workflow engine. Now you can build that directly into your ComfyUI workflows. Click th link for the workflows 👇 Gemini Omni Flash: Text to Video: https://lnkd.in/dbVWR-Qr Gemini Omni Flash: Video Edit: https://lnkd.in/d5PUrEH9 Gemini Omni Flash: Image to Video: https://lnkd.in/dJ8qFXfJ

One of the biggest things that stands out about Comfy MCP is how natural the creative process feels. Instead of thinking about workflows, models, or nodes, you simply describe what you want to create. This showcase demonstrates how a single conversation can generate branded concepts, refine them with feedback, and continue building on the same creative direction—all without ever opening a node graph. It feels less like running workflows and more like collaborating with AI that understands the entire ComfyUI ecosystem. Want to try this workflow yourself? Link in the comments below. https://lnkd.in/g8E2xuxv

We just launched Comfy MCP in public beta, the first MCP built for production pipelines. Connect Claude, Codex, Cursor, or Hermes to the entire ComfyUI ecosystem. → Run any workflow in natural language. Generate 20 ad variants from one hero shot. → Turn storyboard frames into cinematic clips. Build consistent character art across 12 poses. → Re-run your saved product-shot workflow on 5 new images without touching a single node. Every image, video, 3D, and audio model. Hundreds of pre-built workflows. Auto-updated best practices so your agent always knows what's current. Share a workflow URL - your teammate's agent picks it up and runs it too. Same pipeline, fully reproducible, built for long-term projects not one-off generations. No node graphs. No GPU setup. No downloads. No learning curve. Your agent is now a creative technologist. Learn more with the blog: https://lnkd.in/g8E2xuxv #comfyui #comfymcp

ComfyUI Product Drop | Desktop, MCP & VRAM Improvements — Live Demo & Q&A Hosts: Purz & Allyson Your front-row seat to everything we shipped in June! Hear from Comfy's CEO Yoland Yan and product leaders Deep Mehta, Alexis Rolland, Jo Zhang, and Matt Miller who will walk you through what's new and answer questions live. What we'll cover: — ComfyUI Desktop — Core Engine improvements — MCP + waitlist — Team Plans preview Drop your questions in chat and we'll answer them live! ⏰ Subscribe and enable notifications to catch the stream live 💬 Submit your questions in chat during the broadcast or before the stream in the comments! 👍 Leave a like if you'd like to see more content on open source AI! Comfy.org https://comfy.org/ https://lnkd.in/g9g5CwcY

June Launches + Live Q&A Hear from Comfy's CEO Yoland Y. and product leaders Deep M., Alexis ROLLAND, Jo Zhang , and Matt Miller 🤖 Miller who will walk you through what's new and answer questions live. What we'll cover: → ComfyUl Desktop → Core Engine improvements → MCP + waitlist → Team Plans sneak peek Mark your calendars and tune in with the link below 👇 https://lnkd.in/gsYcWkdC

This workflow explores a technique currently gaining attention in AI video: using simple Blender animations as the motion foundation, then bringing them into Seedance 2.0 to add style, and detail from a reference image. Instead of trying to generate motion entirely from scratch, the Blender animation provides structure and control. Seedance 2.0 then interprets that motion while pushing the final result closer to the desired art direction. It is a useful approach for creators who want more consistency, clearer movement, and better control over stylized AI video outputs. Want to try this workflow yourself? Link in the comments below. https://lnkd.in/g7CWD-yV

Seedance 2.0 makes it easy to explore multiple visual directions from the same animation by using reference images to guide the style while preserving the underlying motion. This example begins with a simple Blender scene and basic camera animation, but the approach isn't specific to Blender. The goal is to demonstrate how Seedance, running in ComfyUI, can quickly generate different stylistic interpretations of the same sequence, making it easier to compare ideas and iterate on a scene's overall look. Whether it's used for concept exploration, previs, or creative experimentation, the ability to swap visual styles without changing the animation opens up a fast and flexible way to evaluate different artistic directions. Want to try this workflow yourself? Link in the comments below. https://lnkd.in/g7CWD-yV